Preventable dementia in the Netherlands; population attributable fraction of modifiable risk factors
Abstract
Background: Dementia is a growing public health challenge driven by ageing. A substantial proportion is linked to modifiable cardiovascular, lifestyle, educational, and social factors. International studies suggest that addressing exposures such as hypertension, smoking, physical inactivity, and low education could prevent a meaningful share of cases, although estimates vary. Effective prevention requires understanding which risk factors are most prevalent, how they cluster, and how they vary by age, sex, education, and region.
Methods: We used Lifelines Cohort data (n≈75,000) across multiple follow-up waves to estimate dementia prevalence attributable to 16 modifiable risk factors. Relative risks were derived from meta-analyses. Population attributable fractions (PAFs) were calculated and adjusted for overlap using communality from tetrachoric correlations, yielding weighted PAFs. Analyses were stratified by age, sex, education, and region.
Results: Overall, 50.5% of dementia cases were attributable to the 16 modifiable risk factors. The highest weighted PAFs were unhealthy diet (9.0%), hearing loss (8.8%), physical inactivity (8.6%), social isolation (8.4%), and hypertension (8.3%). Men showed higher PAFs than women for hypertension (9.9% vs 7.0%), hearing loss (10.7% vs 7.4%), and unhealthy diet (10.8% vs 7.5%). In women, poor sleep quality was about twice as high and low cognitive activity about three times higher than in men. Total weighted PAF was higher in low education (54.2%) than high education (41.9%). Regional variation ranged from 47.4% to 55.2%.
Conclusions: These findings align with international estimates (38–56%). Hypertension and physical inactivity remain key drivers, alongside hearing loss, diet, and social isolation. The results support multi-domain prevention strategies targeting cardiovascular, lifestyle, sensory, and social factors, highlighting substantial potential to reduce dementia incidence through integrated population-based prevention.
Key messages: • Not all dementia risk factors are causal; stronger causal evidence (trials, cohorts, MR, quasi-experiments) is needed, considering clustering and subgroup differences for prevention.
• Dementia risk varies by age, sex, education, and region; tailored, context-specific prevention strategies are likely more effective than one-size-fits-all approaches.
Related
- Heat-related mortality attributable to climate change across age and sex in the Netherlands, 1971--2019
- Comparative analysis of plasma pTau217, NfL and eight modifiable risk scores for 11-year dementia prediction in the HUNT Study
- Burden of Dementia attributable to risk factors in Türkiye, 1990–2023
- Population burden of diabetes, hypertension, and cardiovascular disease attributable to obesity: a community-based study in Türkiye
- Patterns of modifiable lifestyle, behavioural and health-related risk factors and their associations with incident dementia in UK Biobank
BibTeX
@article{wenzler2026preventable,
title = {{Preventable dementia in the Netherlands; population attributable fraction of modifiable risk factors}},
author = {AN Wenzler and AC Liefbroer and RC Oude Voshaar and N Smidt},
journal = {The European Journal of Public Health},
year = {2026},
month = oct,
doi = {10.1093/eurpub/ckag169.921},
url = {https://doi.org/10.1093/eurpub/ckag169.921},
note = {CC BY-NC 4.0},
}
RIS
TY - JOUR
TI - Preventable dementia in the Netherlands; population attributable fraction of modifiable risk factors
AU - AN Wenzler
AU - AC Liefbroer
AU - RC Oude Voshaar
AU - N Smidt
PY - 2026
DA - 2026/10/01
JO - The European Journal of Public Health
DO - 10.1093/eurpub/ckag169.921
UR - https://doi.org/10.1093/eurpub/ckag169.921
AB - Background: Dementia is a growing public health challenge driven by ageing. A substantial proportion is linked to modifiable cardiovascular, lifestyle, educational, and social factors. International studies suggest that addressing exposures such as hypertension, smoking, physical inactivity, and low education could prevent a meaningful share of cases, although estimates vary. Effective prevention requires understanding which risk factors are most prevalent, how they cluster, and how they vary by age, sex, education, and region. Methods: We used Lifelines Cohort data (n≈75,000) across multiple follow-up waves to estimate dementia prevalence attributable to 16 modifiable risk factors. Relative risks were derived from meta-analyses. Population attributable fractions (PAFs) were calculated and adjusted for overlap using communality from tetrachoric correlations, yielding weighted PAFs. Analyses were stratified by age, sex, education, and region. Results: Overall, 50.5% of dementia cases were attributable to the 16 modifiable risk factors. The highest weighted PAFs were unhealthy diet (9.0%), hearing loss (8.8%), physical inactivity (8.6%), social isolation (8.4%), and hypertension (8.3%). Men showed higher PAFs than women for hypertension (9.9% vs 7.0%), hearing loss (10.7% vs 7.4%), and unhealthy diet (10.8% vs 7.5%). In women, poor sleep quality was about twice as high and low cognitive activity about three times higher than in men. Total weighted PAF was higher in low education (54.2%) than high education (41.9%). Regional variation ranged from 47.4% to 55.2%. Conclusions: These findings align with international estimates (38–56%). Hypertension and physical inactivity remain key drivers, alongside hearing loss, diet, and social isolation. The results support multi-domain prevention strategies targeting cardiovascular, lifestyle, sensory, and social factors, highlighting substantial potential to reduce dementia incidence through integrated population-based prevention. Key messages: • Not all dementia risk factors are causal; stronger causal evidence (trials, cohorts, MR, quasi-experiments) is needed, considering clustering and subgroup differences for prevention. • Dementia risk varies by age, sex, education, and region; tailored, context-specific prevention strategies are likely more effective than one-size-fits-all approaches.
SN - 1101-1262
SN - 1464-360X
ER -